How reliable is a protein test for measuring your biological age?
A protein test is a promising and now reasonably well-supported way to measure biological ageing, particularly as a predictor of disease and mortality, but its translation into personal use is still very much in development.
A model based on 204 plasma proteins predicts your chronological age with a correlation of 0.94, which amounts to a very close relationship. That accuracy remained stable in large groups of participants in the United Kingdom, China and Finland. This is impressive for a blood test, though a high correlation with your birth year says nothing about whether such a measurement also tells you how healthy you actually are.
The more interesting question is whether protein clocks also say something about disease and mortality. They do: people whose protein clock indicates a higher age than their actual age demonstrably have a greater chance of mortality and of 18 chronic conditions, ranging from heart disease and diabetes to cancer and neurodegeneration. They also correlate with measurable signs of ageing such as telomere length and frailty. These are associations found in large groups followed for an average of more than thirteen years, but cause and effect have not been proven by that alone.
A striking finding is that plasma proteins can also be analysed by organ. Accelerated ageing of the heart is associated with a 250% higher risk of heart failure. Accelerated brain and vascular ageing detectable in the blood can predict Alzheimer's progression just as well as the best current blood marker for that purpose. Roughly one in five people turns out to be ageing at a strongly accelerated rate in one specific organ, while appearing normal in the rest. That detail is lost if you look only at a single overall biological age score.
Reliability therefore depends strongly on what you want to measure. For estimating chronological age, the accuracy is high. As a predictor of health outcomes, the associations are clear and consistent, but have not yet been proven causal. Research also points to non-linear ageing: there are biological peaks of accelerated change around the ages of 41, 57, 60, 67, 70 and 78 years. This may make a one-off measurement less informative than measurements taken at multiple points in time.
Compared with older methods such as the epigenetic clock (based on chemical tags on DNA), protein clocks are a more recent development. A review study from 2017 rated the epigenetic clock as better supported at that time, but since then the evidence for plasma proteins has grown substantially. The two methods measure partly overlapping and partly distinct aspects of ageing, and it has not yet been established which clock is best for which application.
This answer draws on several large studies in which tens of thousands of people were followed for many years, supplemented by smaller studies examining specific organs or the brain. The associations between protein clocks and disease have been found repeatedly in independent groups, which strengthens the findings. Causal evidence is scarce: for only a handful of proteins (including a protein involved in brain ageing and five proteins that influence general ageing) does genetic research support a possible causal role. The comparison with epigenetic clocks comes from a review study from 2017 and is therefore the oldest source in this answer; the field has changed substantially since then.